play nicer with parameters on logs upgraded from 1.4.x
- allows updates for old logs whenever we can (i.e. ok for mdc, not for osc)
- allows updates for everyone if --writeconf is specified at tunefs upgrade
time (requires entire fs to be stopped though)
- allows updates of failover partners if --writeconf is specified
- fixes various other little bugs and error messages
- adds a couple of regression tests for upgrading from 1.4.6

the config borrowed from b_release_1_4_9

r=scjody,vitaly
a fix for extra suffixes for kernel version string which
confuse lbuild.

r=adilger
set external journal device read-only

r=Andreas Dilger

Description: Addition of EXT3_FEATURE_RO_COMPAT_DIR_NLINKS flag for
             > 32000 subdirectories

Details    : Add EXT3_FEATURE_RO_COMPAT_DIR_NLINK flag to
             EXT3_FEATURE_RO_COMPAT_SUPP. This flag will be set whenever
             subdirectory count crosses 32000. This will aid e2fsck to
             correctly handle more than 32000 subdirectories.

- Add some progress indication.
- Fix "cycles" (rm -f erases test dir, so must recreate)

Remove trailing whitespace.

Use lprocfs_rd_atomic() because ns_locks is atomic in HEAD.  Otherwise we get
bogus values from /proc/fs/lustre/ldlm/namespaces/*/lock_count.

long.  Let's declare write_bytes as unsigned long long.

We don't support LUSTRE_KERNEL_VERSION < 45 anymore.

b=11471

r=adilger
Don't print regular error messages about the MGS when we're connecting to
an old MDT.

missing symbol

    OST_CONNECTs to complete at startup) on HEAD

Bugzilla   : 11684
Description: Added D_NETTRACE for recording network packet history
	     (initially only for ptllnd).  Also a separate userspace
	     ptllnd facility to gather history which should really be
	     covered by D_NETTRACE too, if only CDEBUG recorded history in
	     userspace.

Avoid synchronous disconnects for liblustre clients, already in Cray
upstream.
b=10803
r=green

Description: large O_DIRECT requests fail under memory pressure/fragmentation
Details    : Large single O_DIRECT read and write calls can fail to allocate
	     a sufficiently large buffer to process the request.  In case of
	     allocation failure the allocation is retried with a smaller
	     buffer and broken into smaller requests.
b=10901
r=vitaly r=green

Remove warning about non-portable initializer.

Reorder include files so that the cookie check doesn't always fail.

b=11471

(borrowed from b1_4).  the tcp-zerocopy-2.6-sles10.patch patch removed for
fixing an lbuild failure and because lnet does not need that patch anymore.

b=10947

r=green
don't allocate inside a spinlock

b=11248

b=11248
r=adilger

b=11471

b=9829
r=adilger

Description: client incorrectly hits assertion in ptlrpc_replay_req()
Details    : for a short time RPCs with bulk IO are in the replay list,
       but replay of bulk IOs is unimplemented.  If the OST filesystem
       is corrupted due to disk cache incoherency and then replay is
       started it is possible to trip an assertion.  Avoid putting
       committed RPCs into the replay list at all to avoid this issue.

Some minor cleanups to make extents, mballoc kernel patches the same.
Use ext3_fsblk_t as appropriate for 2.6.18.
b=11675

Add ChangeLog entry
b=11669

Fix 2.6.18 ext3 mballoc, extents patches for several omissions.
Reorder patch hunks to match other patches to facilitate comparisons.
b=10090  ldiskfs corruption under memory pressure, coverity fixes
b=6191   mballoc fails on x86_64 > 2TB
b=10634  mballoc using wrong find_next_bit() on big endian systems
b=6449   don't print "too long searching" message to console

Compile tested.

Quiet -Wundef compiler warning.

Line wrap comments at 80 columns.

Better comment.

Clarify comment.